Deep learning-based classification methods of hyperspectral images (HSIs) have made significant progress recently, catching the attention of academia and industry; however, the existing studies of classification of HSIs mainly focus on the closed-set environment with the assumption that ground classes are fixed and known, ignoring the complexity and diversity of ground objects in the real world. As a result, the unknown classes will be forced into known classes. To solve this problem, we propose a novel spectral-spatial evidential learning (SSEL) network that combines an improved generative adversarial network (GAN) and evidential theory for open-set classification of HSIs. First, a domain adaptation (DA) strategy is embedded into GAN to generate high-quality samples by reducing the distribution discrepancy between generated and real samples. Second, the discriminator is devised to extract spectral-spatial features and output multiclass evidence for closed-set classification and uncertainty estimation. A new classification function called evidence-based loss is designed for the discriminator to guide the evidence-collection process. Additionally, a novel adversarial objective function is defined, where the discriminator loss is devised to predict real samples belonging to the true class and generated samples belonging to “none of the classes. The generator loss is developed to generate samples consistent with the label category. Finally, the class and corresponding uncertainty can be calculated based on the collected evidence and the appropriate open-set classification of HSIs. Extensive experiments on three benchmark HSIs show that our proposed method achieves competitive performance on closed-set and open-set classification of HSIs compared with existing state-of-the-art methods.

The rapid development of hyperspectral remote sensing technology, has led to an explosion in the number of available hyperspectral images (HSI). The fast and accurate characterization of HSI poses a significant challenge for remote sensing scientists. Currently, deep learning strategies with various neural networks have been successfully applied for HSI classification using the concept of the “dataset-model”. Still, there is a need to develop universal deep learning models for HSI classification using a continual updating strategy. This paper presents a life-long learning strategy to continually update model weights with the help of continual spectral-spatial feature distillation. Specifically, the proposed method introduces a spectral-spatial distillation strategy to retain knowledge of the previous well-trained model. Meanwhile, the learning metric term is integrated into a multi-level feature extraction to minimize the spectral-spatial feature discrepancy between the previous model and the new one. The experimental results indicate that our method achieves superior performance for continual HSI classification tasks without suffering from the persistent loss of characterization memory.

Accurate and timely monitoring of wildfires is crucial for reducing property damage and casualties. In recent years, advances in satellite technology have enabled the comprehensive, timely, and rapid recording of various abrupt events on the Earth’s surface. However, achieving a balance between temporal and spatial resolution remains a significant challenge for remote sensing, hindering the quick and accurate detection of wildfires. This letter proposes a novel framework for the near real-time monitoring of wildfire coupled with the BRDF model and deep learning technology, which enables near real-time detection of wildfire by assessing the degree to which the observed value of geostationary satellite image deviates from the predicted theoretical observation value. The experimental results show that the proposed method is capable of effectively detecting wildfires in near real-time. Moreover, the encouraging results suggest that the method holds promise for monitoring the spread of wildfire to a certain extent.

Unseen category prediction is a common challenge for real-world applications, especially for remote sensing (RS) imagery interpretation. Zero-shot learning (ZSL)-based scene classification methods have made significant progress recently, providing an effective solution for unseen scene recognition with semantic embeddings that link seen and unseen classes in the field of RS. However, existing ZSL methods mainly focus on semantic feature exploration, and they failed to combine image features and semantic features effectively. To address the aforementioned challenges, we propose a novel knowledge graph (KG)-based ZSL model that adeptly integrates both image and semantic features for disaster RS scene recognition. First, we construct an RS KG to generate semantic features of RS scenes, enhancing the reasoning ability from conventional RS scene categories to disaster RS scene categories. Second, we propose an interactive-attention mechanism to integrate image and semantic features, focusing on the most informative regions. Finally, we introduce an RS domain adapter that enables the model to better adapt to RS data, reproject common features into the RS domain, and thus solve zero-shot RS scene classification tasks. To demonstrate the effectiveness of our method, we construct an RS disaster scene dataset, which contains 8700 high-quality disaster scenes. Extensive experiments show that our proposed method outperforms current state-of-the-art (SOTA) methods under zero-shot RS image scene classification settings.

With the development of remote sensing technology, the continuing accumulation of remote sensing data has brought great challenges to the remote sensing field. Although multiple deep-learning-based classification methods have made great progress in scene classification tasks, they are still unable to address the problem of model learning continuously. Facing the constantly updated remote sensing data stream, there is an inevitable problem of forgetting historical information in the model training, which leads to catastrophic forgetting. Therefore, we propose a continual contrastive learning method based on knowledge distillation and contrastive learning in this paper, which is named the Continual Contrastive Learning Network (CCLNet). To overcome the problem of knowledge forgetting, we first designed a knowledge distillation module based on a spatial feature which contains sufficient historical knowledge. The spatial and category-level knowledge distillation enables the model to effectively preserve the already learned knowledge in the current scene classification model. Then, we introduced contrastive learning by leveraging the comparison of augmented samples and minimizing the distance in the feature space to further enhance the extracted feature during the continual learning process. To evaluate the performance of our designed model on streaming remote sensing scene data, we performed three steps of continuous learning experiments on three datasets, the AID, RSI, and NWPU datasets, and simulated the streaming of remote sensing scene data with the aggregate of the three datasets. We also compared other benchmark continual learning models. The experimental results demonstrate that our method achieved superior performance in the continuous scene classification task.

Recently, developments in hyperspectral image (HSI) classification have brought increasing attention to the challenges of the open-set problem. However, current open-set methods generally overlook the intra-class multimodal structure, making it difficult to comprehensively capture the global data distribution, which in turn reduces their ability to distinguish known from unknown classes. To address this, we propose a novel global distribution-aware network (GDAN) that jointly performs pixel-wise HSI classification and trustworthy uncertainty-aware identification of unknown class. First, a generative adversarial network (GAN) is employed as the backbone, enhanced with a self-attention (SA) module to capture long-range dependencies across the extensive spectral bands of hyperspectral data. Second, an interpretable open-set HSI classification framework is designed, combining GAN with Markov Chain Monte Carlo (MCMC) to model global distribution by exploring intra-class multimodal structures and estimate predictive uncertainty. In this framework, the traditionally fixed discriminator weights are reformulated as probability distributions, and posterior inference is conducted using MCMC within a Bayesian framework. Finally, accurate categories and predictive uncertainty of ground objects can be obtained through posterior sampling, while samples with high uncertainty are assigned to the unknown class, thus enabling accurate open-set HSI classification. Extensive experiments on three benchmark HSI datasets demonstrate the superiority of the proposed GDAN for open-set HSI classification, yielding overall classification accuracies of 94.6%, 92.6%, and 94.8% in the 200-sample scenario.